Windows PowerShell is the successor of the windows cmd language, which itself has its roots in the ms-dos Bat language. All recent versions of Windows offer PowerShell (PS). PS may be seen as Microsoft's answer to the shells common in Unix/Linux (such as csh, bash, etc.). Its name implies that Microsoft sees the shell as powerful, which it arguably is.
In these notes some important PS commands are listed and PowerShell's most notable feature, the object pipeline, is discussed. From the outset it is important to note that, in contrast to Linux/Unix, Windows PowerShell is completely case-insensitive.
The monospace text snippets below are valid PS and may be copied, pasted, and executed in a PowerShell- or a PowerShell_ISE-session. This is why the notes form a "Cheatsheet". As is common for cheatsheets, there is hardly any explanation, the examples speak for themselves. It must be stressed here that many of the basic PS commands are not at all orthogonal, so that many variant pipelines can lead to the same effect. Often an example is one out of a multitude of possibilities accomplishing the same task.
The last two sections are about search in and traversal of the Windows Registry by means of PowerShell.
cmd
, powershell.exe
,
or powershell_ise
and a corresponding window opens at the current directory (aka folder). This is
most likely the shortest route to opening a cmd, ps, or ps_ise
session from a given directory.
Control Panel
(in Dutch: Configuratiescherm
), Regedit
, PowerShell
,
or cmd
.
This opens a window with the name of the chosen program (Control Panel, Regedit
, etc).
Click on it once and the program chosen opens.
Control Panel/User Accounts/User Accounts/
. Then left panel:
change my environment variables
. (Choose in Control Panel: view by category
).
Control Panel: view by category
. Then
Control Panel/System and Security/System/Advanced System Settings
.
Button: Environment Variables
. Set in panel system variable
.
Change is persistent, a change (see below) within PowerShell is volatile.
setx
sets persistent user environment variable. To unset:
go to HKCU/environment
(in the Registry) or to Control Panel/User Accounts
.
ISE stands for Integrated Scripting Environment. ISE—a standard part of Windows 10—is an editing and execution environment for PowerShell.
An important advantage of the use of ISE over pure PowerShell, is that ISE offers pop-up help: when the user is typing a command line, ISE often pops up a relevant list of methods and properties. By double clicking on an entry in the pop-up list, the entry is appended to the command line.
A few noticeable points:Microsoft.PowerShellISE_profile.ps1
is executed.
Set-ExecutionPolicy RemoteSigned -Scope CurrentUser
unblock-file .\script.ps1where
script
is the name of the downloaded script (which is in the current directory).
.\script.ps1
, where script
is the name of the script.
$Host.UI.RawUI.BufferSize = New-Object $Host.UI.RawUI.BufferSize.GetType().Fullname (150, 40)(The pure PS console does not allow shrinking of width).
$Host.PrivateData.FontSize = 13(pure PS has a different
$Host.PrivateData
object).
Start-Process PowerShell -Verb runAs
PS C:\Users\myname> Get-ChildItemreturns a collection of objects that belong to the different files in the present folder. Because the command
Get-ChildItem
is the end stage of a (one-stage) pipeline, some selected properties
(namely Mode, LastWriteTime, Length, Name) together with their values are written to screen, one line per file.
The names of members (methods and properties) of an object can be obtained by piping
the output of a cmdlet to Get-Member
(see below).
Any cmdlet parameter (aka flag) can be truncated to the extent that it is still unique.
A few examples of cmdlets and their parameters:
Get-Help # Gets help about a cmdlet, # example: get-help get-help Get-PSdrive # List available PSdrives, such # as c:, env:, hklm:, hkcu:, alias:, etc. Get-ChildItem # In the Registry: children are subkeys # of the current key. Get-ChildItem # In the File System: children are subfolders and filenames # of the current folder. Get-ChildItem -recurse # Lists recursively all children # of current PSdrive, Folder, or Registry key. Get-ChildItem -rec -force # Include hidden folders # (flag -hidden searches hidden directories only) (Get-ChildItem).name # List names ('name' is an object property) # of files and directories in current folder. Get-ChildItem -name # Equivalent to (Get-ChildItem).name # (Get-ChildItem).count # Number of entries in the collection of objects returned by Get-ChildItem # (count is a property of such a collection).
c:, env:, alias:, HKLM:
. Here c:
is the usual Windows c-drive; env:
is the space of Windows environmental
variables; alias:
is the collection of cmdlet aliases; HKLM
is a hive in the Registry.
Standard one enters a PS session in the home folder (home directory) of the user.
To switch a PS session to another PSdrive or folder
and get the children of the new location, proceed as follows:
Switch to env:
Set-Location env: # Prompt character becomes `Env:\>` # (environment variables) Env:\> Get-Childitem # Get all environment variables # Env:\> Get-Childitem userprofile # Get environment variable `userprofile` # (returns with: USERPROFILE C:\Users\user_name)Switch to alias:
Env:\> Set-Location alias: # Prompt character becomes Alias:\> Alias:\> Get-Childitem # All children (i.e., all aliases)Back to default drive:
Alias:\> Set-Location C:\ # Prompt character becomes C:\> # C:\> Set-Location $env:userprofile # Use environment variable `userprofile` to # switch to C:\Users\user_name (home folder of user). C:\Users\user_name>$alias:ls # Get what alias 'ls' stands for # (namely Get-ChildItem)Here
$env:
and $alias:
(note the $ prefixing the names) refer to the PSdrives env:
and alias:
, respectively. Thus, the prefixing by the variables $env:
and $alias:
gives access
to the respective PSdrives without need to actually change to these drives.
IMPORTANT: Cmdlets pass objects through pipelines, not character streams as in Unix.
The pipeline character is |
(ASCII 124). It must be followed by a command that can handle the output passed through the pipeline; usually this is a cmdlet.
Example of a pipeline consisting of three stages:
Get-ChildItem *.txt | Where-Object length -lt 1000 | Sort-Object lengthThis returns a list of the names and properties of files with extension
.txt
in the current folder.
Shown are the .txt
files that have a size of less than 1000 bytes. The list is sorted on file size.
The output of Get-ChildItem flop.txt
is an object and by the pipe
Get-ChildItem flop.txt | Get-Membera full list is obtained of the members (in object-oriented languages known as methods or properties) of the object associated with the file
flop.txt
.
In this list we see that one of the members is named name
.
Members are selected by a dot, as is usual in object-oriented languages. Thus, one can write:
(Get-ChildItem flop.txt).name # -> flop.txtOne of the properties of any file is
LastWriteTime
. This property can be set
to the present date and time without affecting the
content of the file (cf. touch
in Unix):
(Get-ChildItem file.txt).LastWriteTime = Get-Date # "touch" file.txt
The object created by a cmdlet depends in general on the cmdlet's parameters (flags).
For example, by adding the flag -name
(Get-ChildItem flop.txt -name).name # -> nullwe get an empty result. Inspection by
Get-Member
shows that indeed the object created by Get-ChildItem flop.txt -name
does not posses a member by the name name
.
The object that is passed through a pipeline is referred to by the automatic variable $_
,
which may be used only inside a script block. (A script block is a collection of statements enclosed in curly brackets.)
Accordingly, a member named "member_name
" of the object passed is referred to as $_.member_name
.
The cmdlet Rename-Item
may be used to change file names and file extensions. The old
file name can be entered through a pipeline. The new name follows the parameter -newname
"flop.txt" | Rename-Item -newname flap.txt # We now have a file named flap.txt in the folderInvoking the automatic variable, we perform the following trivial renaming:
Get-ChildItem flap.txt | Rename-Item -new {$_.name} # Renamed flap.txt to flap.txtHowever, if the piped object
$_
does not have the member (property) name
, an
error occurs:
Get-ChildItem flap.txt -name | Rename-Item -new {$_.name} # Error: parameter $_.name is null
When more than one object is outputted by a cmdlet in the pipeline, these objects are first stored in a temporary buffer. After the buffer is filled, the cmdlet in the next stage performs its task by looping over the buffer and reading the objects one by one. When this stage is not the last of the pipeline, the cmdlet puts its outputted objects in yet another temporary buffer that serves as the input for the next stage. An object that drops out at the end of the pipeline is usually written to screen by a screen writing method of the object. The concept of intermediate buffering is very important for the understanding of PowerShell pipelines.
For example in the following statement, Get-ChildItem
fills a temporary buffer with
objects that all have the property basename
(which is also an object).
The cmdlet Select-Object
selects the basename properties and stores them in a yet another temporary buffer, which is passed to Sort-Object
, which sorts the elements of the buffer and writes the names of the basename objects, i.e., Sort-Object
writes the file names without file extension.
Get-ChildItem | Select-Object basename | Sort-Object *The result on the screen is a list of the names of all files in the current folder. The list is in alphabetic order.
The use (or missing) of parameters of cmdlets that receive input in pipelines can generate unexpected errors. Example:
Move-Item *.txt subdirectorymoves all
.txt
files to folder subdirectory
. Analogously, one could assume that the following
pipe would move the files one folder up:
Get-ChildItem *.txt | Move-Item ..\ # Error!This gives the error message that
Move-Item
lacks input:
Move-Item : The input object cannot be bound to any parameters for the command either because the command does not take pipeline input or the input and its properties do not match any of the parameters that take pipeline input.
-destination
remedies this and
Get-ChildItem *.txt | Move-Item -destination ..\moves without complaint all
.txt
files one level up.
ac = Add-Content # Example: ac -value 'The End' -path 'flop.txt' # (appends value to file) cat = gc = type = Get-Content # Get the content of a file; # returns an array with one line per element cd = sl = Set-Location # Change folder, Registry key, or PSdrive. # Example: cd env:, cd HKLM: cls = clear = Clear-Host # Clears console # del = erase = Remove-Item # Remove files, registry keys, etc. # dir = gci = ls = Get-Childitem # List children in current PSdrive, folder, Registry key # echo = write = Write-Output # String to output array. Array is sent to console, into # pipeline, or redirected/appended to file foreach = % = Foreach-Object # Only in pipeline: for each object crossing the pipeline # Do not confuse with language construct of the same name ft = Format-Table # Example: ls *.jpg |ft directory, length, name -AutoSize -Wrap # fl = Format-List # Example: ls env:Path |fl # (gives wrapped output of environment variable "Path") gal = Get-Alias # "Get-Alias -definition cmdlet", gives aliases of cmdlet # "Get-Alias [-name] alias", gives name of cmdlet called by alias gcm = Get-Command # Get all commands (cmdlets, functions, and aliases). # gcm -CommandType Alias -> all aliases gm = Get-Member # Example: ls flop.txt | gm # (all members of object flop.txt) gp = Get-ItemProperty # In file system: gp * gives same output as ls * # In Registry: value entries (names and values) gpv = Get-ItemPropertyValue # In filesystem: get prop's of files. Ex: gpv *.txt basename (names of .txt files) # Ex: In HKCU:\SOFTWARE\Microsoft\Accessibility: gpv -name cursorsize (returns number) gv = Get-Variable # Get names and values of # all session variables ni = New-Item # Create new file, directory, symbolic link, # registry key, or registry entry ps = gps = Get-Process # List running processes. # pwd = gl = Get-Location # Current directory (folder) # or Registry key ren = rni = Rename-Item # Examples: ren report.doc report.txt # and: ls report.doc | ren -newname report.txt rmdir = rm = ri = Remove-Item # Remove directories, files, registry keys, etc. # rv = Remove-Variable # Remove variable (name without $ prefix, while # note that variable names must begin with $) select = Select-Object # Select specified properties of piped object # Example: ps |select Processname | select -first 10 sleep = Start-Sleep # Sleep -sec 1 # (sleep 1 second) sls = Select-String # Example: sls foo.txt -patt '^\S' (a regular expression # giving all lines that do not start with blank, tab, or EOL) where = ? = Where-Object # Only in pipeline. # Example: ls -recurse |? name -like '*Pict*' $env:userprofile = ~ # Example: cd ~ (change folder to home folder of user).
C:\Program Files
, for example. It outputs
the names of .dll
files of size less than 10000 bytes in the current folder and
all its subfolders:
Get-ChildItem -recurse -path *.dll | Where-Object {$_.length -lt 10000} | Sort-Object -property Length | Format-Table -property name, length, directory -wrapThe parameter
-path
, being default, can be omitted, as can the parameter -property
in two of the stages.
The command can be shortened further by introducing aliases
and abbreviated parameters. For clarity, the statement is split by assigning an array object to the variable
$a
. (Recall here that variable names are text strings that begin with a dollar sign). Note that
the cmdlet Where-Object
(alias: ?) can recognize names of object properties without use of a script block.
That is, {$_.length -lt 10000}
is equivalent to length -lt 10000
. Thus,
$a = ls -r *.dll |? length -lt 10000 # Store in $a all .dll files from current directory downward # with file sizes < 10000 bytesThe array object
$a
is piped to the alias sort
of the cmdlet Sort-Object
and the sorted object goes to Format-Table
:
$a | sort length | ft name, length, directory -w # Sort entries of array $a on file size (length) and tabulate # formatted name, length, and directoryFinally, in one statement:
ls -r *.dll |? length -lt 10000 |sort length |ft name, length, directory -w
"<"
operator is reserved for future use, so that the cmd-mode/Unix redirection:
./program.exe < input.txtdoes not work. Instead, pipe the input:
cat input.txt | ./program.exeThe
">"
operator redirects to output and ">>"
appends to output, just like they do in cmd-mode and Unix.
PSnotes.txt
:
ls PSnotes.txt |fl * # Lists all properties: Directory, LastAccessTime, Basename, etc. ls PSnotes.txt |fl LastAccessTime, Basename # Lists two properties: date/time of last access and file name.If you want to tabulate (instead of list) names and values of one or more properties then pipe to
ft
. Example:
ls PSnotes.txt |ft LastAccessTime, Basename # Tabulates date and time of last access and file nameThe difference between tabulating (ft) and listing (fl) properties is minor.
To list the content of the lines in foo.out
that begin with at least four spaces together with their sequence numbers
use select-string
(alias sls
). The cmdlet has the parameter -pattern
that specifies a regexp:
sls foo.out -pattern '^[ ]{4,}' | ft linenumber, line(Note that an empty line may not contain spaces and is then not shown by this command).
was
' is found in all .txt
files in the present directory (folder)
by application of sls
:
sls *.txt -patt 'was' |ft -wrap filename, linenumber, lineHere
-wrap
indicates that line
is not truncated but wrapped.
Find all directories called winx
from the present directory downward (-rec
). Inspect also
hidden directories (-force
) and suppress error messages (-ea 0
):
ls winx -dir -rec -force -ea 0 |ft
Where-Object
, which only appears in pipelines, has alias "?".
As an example, list the basenames ending with the letter r
in the current directory.
Use the operator -match
with a regexp and recall that the symbol $
indicates the end of a string:
ls |? basename -match 'r$'This lists the basenames ending on 'r' plus additional information (mode, write time, length, full file name). If only the basename is to be listed, use:
ls |? basename -match 'r$' |ft basename
Example of the comparison operators: -in, -notmatch, -and, -notlike
.
The first statement below limits the list of 'svchost' and 'firefox' processes to the first 10 in total.
The next statement uses a script block, which is
the code snippet between curly brackets:
ps |? ProcessName -in "svchost", "firefox" | select -f 10 |ft processname, PagedMemorySize ls |? {$_.name -notmatch 'e$' -and $_.name -notlike 'c*'} # Names not ending on "e" (regexp) or beginning with "c"
%
is an alias for ForEach-Object
, as is foreach
. Example:
Get-Alias |% {if ($_.name -match '^s') {Write-Host $_.name ' stands for: ' $_.definition}}This gives:
sajb stands for: Start-Job ... si stands for: Set-Item ... sleep stands for: Start-Sleep ... swmi stands for: Set-WMIInstance
Get-Alias
sends 158 objects (aliases) through the pipeline. All objects have a property name
,
which is matched against a regexp. The regexp checks if the first letter is 's' or 'S'. If true, the object attribute
name
, the string "stands for:", and the object property definition
are written.
The Write-Host
cmdlet, which allows a certain freedom in the
format of its output, writes to a host—which in an interactive session is the screen.
A boolean expression must be surrounded by round brackets,
the body of the Foreach-Object must be enclosed by curly brackets (the outer ones), and the body of the
true branch is within curly brackets too.
Note that a list of aliases starting with 's' or 'S' can be obtained in a prescribed format
from the shorter Where-Object (?)
statements:
Get-Alias |? name -match '^s' # regexp Get-alias |? name -like 's*' # string + wildcardAnd still shorter:
gal s*
ForEach-Object
loop; statements are separated by semicolons.
Example: list the names and count the number of lines of the .txt
files in the present folder:
ls *.txt |% {Write-Host $_.name, " " -NoNewline; (cat $_).count }Here
-NoNewLine
is a parameter of Write-Host
. Remember that cat
returns an array object that has the method count
.
As yet another example, notice first that
gal cat |? displaynamedisplays the string
Alias cat -> Get-ContentThe verb "Get" can be captured from this string by the comparison operator
-match
with the regexp (note the parentheses indicating a capture):
$reg = '-> (\w*)-'and the captured result appears in
$matches[1]
. Here \w*
indicates an
arbitrary number of word characters.
The alias gal (Get-Alias)
, without arguments, fills a buffer containing all aliases.
Each object in the buffer has the property "displayname".
Hence to get a sorted list of the unique verbs of all aliases, issue:
gal|? displayname -match '-> (\w*)-' |% {$matches[1]} |sort -uniqueThree aliases are missed here (namely
ise, man, md
) because their definition lacks the hyphen: -
.
An example of foreach
as a language construct:
$letterArray = "a","b","c","d" foreach($letter in $letterArray){ Write-Host -ForeGroundColor green $letter }Yet another example as language construct:
$ff = ps firefox # Usually there is more than one firefox process active foreach($p in $ff) {$p.starttime}This gives something like (in Dutch):
zaterdag 8 juni 2019 07:47:44 zaterdag 8 juni 2019 13:53:41 zaterdag 8 juni 2019 07:47:30 zaterdag 8 juni 2019 07:47:28 zaterdag 8 juni 2019 15:21:39 zaterdag 8 juni 2019 13:42:46 zaterdag 8 juni 2019 11:52:45Incidentally, the very same output is obtained by:
(ps firefox).starttime
Select-String
(alias sls
) above. Because it
is an extremely useful tool—and not an easy one as it depends heavily on regular
expressions—we give three more examples of its use.
Consider as a first example the statements
$address = 'https://131.174.138.39/~pwormer/teachmat/PS_cheat_sheet.html' $parsed = $address |sls -patt 'https://([0-9.]*)/(.*)$'The second statement breaks out the IP address and the path from
$address
and assigns them to submembers of the object
$parsed
. This object has several members, one of them matches
. The one-stage pipeline:
$parsed.matcheslists on the screen the names and values of all properties of the object
$parsed.matches
. It gives:
Groups : {0, 1, 2} Success : True Name : 0 Captures : {0} Index : 0 Length : 60 Value : https://131.174.138.39/~pwormer/teachmat/PS_cheat_sheet.htmlThe array
$parsed.matches.groups
contains objects that in turn contain the captures, i.e., the subexpressions in the regular expression that are between parentheses. As always the zeroth element gives the total string (as does the property $parsed.matches.Value
). In summary,
$parsed.matches.groups[0].value : https://131.174.138.39/~pwormer/teachmat/PS_cheat_sheet.html $parsed.matches.groups[1].value : 131.174.138.39 $parsed.matches.groups[2].value : ~pwormer/teachmat/PS_cheat_sheet.html
As another example, we remark that the present file, called PS_cheat_sheet.html
, contains strings that are enclosed
in <b> and </b>, i.e., the strings are displayed in bold face. We will now construct a PS pipeline that lists these strings.
(The pipeline does not work correctly when bold text extends over different lines. Moreover,
the search will be non-greedy, which means that only the first bold string in each line is returned).
First we introduce a regular expression (regexp) that captures bold HTML text:
$reg = '<b>(.*?)</b>'This regexp matches the first arbitrary string in a line that is enclosed within
<b>
and </b>
.
It captures the string matching the regexp inside the parentheses. The latter regexp matches zero or more (*
) arbitrary characters (.
) in a non-greedy search (?
).
The regexp $reg
is used in:
sls PS_cheat_sheet.html -patt $reg |% {$_.matches.groups[1].value}which writes all the bold strings in the present notes.
As a final example we consider the file foo.txt
with contents:
In columns 20...25 (six columns) are positive, negative, and unsigned integers. 101 xxxxx -23 ddd 20 ddd +2 eee 30 %^&&fghu -100 ffff 40 qawer 1000 fffqq And yet one more 1 unsigned number in column 23!The following statements list the sum (=880) of the numbers in columns 20-25:
$s=0; (sls foo.txt -patt '.{19}([0-9-+]{1,6})').matches |% {$s+= $_.groups[1].value}; $sThe regexp skips arbitrary characters in the first 19 columns and captures digits and plus/minus signs in columns 20-25. The sum is accumulated in
$s
where PowerShell converts
the captured strings to integers.
New-Item
(alias ni
) creates a new item. An item is
one of the following:
manual
. (In Dutch Desktop is Bureablad).
Both PowerShell and cmd require admin privilege to do this.
The newly created link manual
targets the file manual.txt
that exists in the current folder:
New-Item -ItemType SymbolicLink -Path C:\Users\paul\Bureaublad\manual -Target .\manual.txtThis can be shortened to:
ni -ty sym -p C:\Users\paul\Bureaublad\manual -v manual.txtThe parameter
Target
has the alias Value
, ItemType
has the alias Type
,
and the current folder indicator .\
may be omitted.
Create more than one .txt
file at once, example:
'file1', 'file2', 'file3' |% {ni $_'.txt'}Also more than one folder can be created in this way:
'dir1', 'dir2', 'dir3' |% {ni -ty dir $_}
Rename-Item
(aliases: ren
and rni
) renames items such as
files, folders, and registry keys. In contrast to the cmd command rename
, it does not allow
a wildcard in the name of the files. Although
ren report.txt report.docis correct in cmd-mode as well as in PowerShell, the command that includes the wildcard
*
ren *.txt *.doc # Error in PS!only works in cmd-mode. PowerShell returns an error.
The cmdlet rename-item
can take piped input for the old name and recognizes
the flag -newname
for the new name. For example,
'report.txt' | ren -newname report.docgives the required change of the file extension.
To change multiple names at once, one may use the -replace
operator. Its syntax is:
string -replace regexp, new_name
In string
every substring that matches the regular expression
regexp
is replaced by new_name
. For example,
'report.txt' -replace '\.txt$', '.doc' # -> 'report.doc'The regular expression
'\.txt$'
is anchored at the end of the string by "$
" and
the dot is escaped so that its meaning is not the regexp arbitrary character but the ordinary punctuation mark.
Thus,
ls *.txt | ren -new { $_.name -replace '\.txt$','.doc' }replaces in the current directory all file extensions
.txt
by .doc
.
The value of the script block (the part between curly brackets) is a string: the new file name.
In the previous example the PS statement is considerably more difficult to memorize than
the equivalent cmd statement. Because that happens more often, it is
useful to mention that cmd /c
invokes a cmd command
from within Powershell (the flag /c
stands for "command"). Thus,
cmd /c rename *.doc *.pdfchanges the extension
.doc
to .pdf
for all items in the current folder.
Parenthetically, the following PS command achieves the same task without invoking -replace
:
(ls *.doc).basename |% {ren .\$_.doc .\$_.pdf}This statement is not easy to memorize either.
Note. The text with hyperlink
-replace
operator
states explicitly that the first argument of -replace
must be a regular expression.
In the fourth example of
get-help rename-item -examplesthe first argument of
-replace
is '.txt'
. At first sight this looks like an ordinary string,
but if we remember that the dot stands for an arbitrary character, we do recognize it as a regexp.
[int], [long], [string], [char], [bool], [byte], [double], [decimal], [single], [array], [xml],
[hashtable], [PSCustomObject]
.
[string]$s = 'Peanuts' # Explicit declaration $s = 4 # Assign new string '4' $s * 3 # Gives 444 (triplicates the string) [int]$s = 4 # Explicit change of type (to int32) $s * 3 # Gives 12 $s = 'bear' # Error --> Cannot convert value "bear" to type "System.Int32" [string]$s = 'bear' # Explicit recasting is OK $s = [char]0x07 # Recasting is OK on RHS (hex-char 0x07 sounds "Bell", warning sound)Another type casting example:
'a', 'b', 'c', 'd' > letters.txt # Each letter on a line in file letters.txt. Output in UTF-16 (!) [string]$letter = cat letters.txt # Get-Content, assign to string $letter # --> a b c d (a single string) $letter = [string](cat letters.txt) # The same. Brackets are necessary to cast the output of the cmdletUse
set
to assign a constant:
set pi 3.14 -opt const # Assign constant; (No $ in set pi !) $pi # Gives 3.14 (Use $ in reference to the constant just set without $!) $pi = 2 # -> Error: Cannot overwrite variable pi because it is read-only or constant. set pie $([math]::pi) -opt const $pie # -> 3.14159265358979Arrays
$a = @('a', 'b') # Array (round brackets, colons as separators) $a += 'c' # Push 'c' $a += 'd' # Push 'd' $a # --> a\n b\n c\n d (vertical stack) $a.GetType() # --> True True Object[] System.Array $arry = 1,2,3,4,5 # Array, $arry[0] is 1, $arry[$arry.length-1] is 5. $arry.GetType() # --> True True Object[] System.ArrayAssociative arrays (aka dictionaries or hash tables)
$assoc = @{one=1 ; two=2} # Associative array (curly brackets, semicolons as separators) $assoc.one # -> 1 $assoc['two'] # -> 2 $assoc['three'] = 3 # Adds a member $assoc.four = 4 # Adds a member $assoc.GetType() # --> True True Hashtable System.Object $assoc #The last statement (
$assoc
) gives:
Name Value ---- ----- four 4 one 1 three 3 two 2
$
are evaluated. For example,
$five = 5 '3*$five' -> 3*$five "3*$five" -> 3*5 "$(3*5)" -> 15Backtick escapes under double quotes. For example, escaping the dollar symbol:
"`$(3*5)"
gives $(3*5)
.
Backticks are unchanged under single quotes:
'`$(3*5)'
gives '`$(3*5)'
.
"`n"
gives the newline character.
"John Doe"
can be appended to a file simply by "John Doe" >> out.txt
.
The file out.txt
will be in UTF-16.
add-content
(alias ac
) writes to file
by default in ANSI (Windows-1252), and allows specification of other encodings.
Here-strings
@' ... '@
(no expansion) and @" ... "@
(with expansion).
Note that the openings @'
and @"
must start in column 1 and be on a single line. The same holds for the endings '@
and "@
.
$a -eq "Big Brother"
:
@' $a $(3*5) '@gives
$a $(3*5)while
@" $a $(3*5) "@gives
Big Brother 15
C#
.
The format operator is -f
. Schematically it is used like:
"String containing format-items" -f elements to be formatted separated by commasExamples:
"{0, 0:f4} is rounded to 4 decimals" -f 12.34567 # -> 12.3457 is rounded to 4 decimals "12345 converted to hex is: {0,10:x}" -f 12345 # -> 12345 converted to hex is: 3039 "Prices are {0, 0:c} and {1, 8:c0}" -f 12.3998, 1.99 # -> Prices are € 12,40 and € 2On the left-hand-side of the operator
-f
there is a string containing format-items:
{index[,alignment]:[formatString[number]]}[Square brackets surround optional values]. The index is 0-based and refers to the position in the array on the right of
-f
.
The alignment gives the total number of spaces reserved for the output, with the output string being right-aligned
in the reserved space. If the alignment is too small, the system takes the spaces it needs.
After the colon is the format string (see below).
In the case of floating point numbers, the number of decimals is given by the number after the format string.
By default it is 2. The number of digits before the decimal point is adapted by the system so as not to lose information.
Write-Host
:
$str = "One decimal:{1,5:n1}; two decimals:{0, 7:n2}; three decimals:{2, 10:n3}" Write-Host($str -f 2.141, 1.141, 3.141)Note the order: second argument first, then the first, and finally the third. Also look at the spacings dictated by the alignment parameter:
One decimal: 1.1; two decimals: 2.14; three decimals: 3.141 ||||| ||||||| ||||||||||The following is an (incomplete) list of format strings, optionally they may be appended by an integer giving the number of decimals:
$, €, ...
)
n:fd
gives field of length n
and d
digits after the decimal point. If n
is too small the system adapts.
gp = Get-ItemProperty
):
gp *.ps1 |% {$_.length}A script block allows computation (
1kb
is a literal constant of value 1024):
gp *.ps1 |% {$_.length/1kb} # Lengths in kilobyteList basename and length in a pretty format (round brackets are needed to give priority to the division):
gp *.ps1 |% { "{0, 31} {1, 6:f1}" -f $_.basename, ($_.length/1kb)}As usual, there is a simple alternative to get (almost) the same info (lengths in bytes):
ls *.ps1 |ft basename, length
"This is a valuable string" |gmwe see lots of string methods among which is
split
.
That is,
$a = ("This is a valuable string").split(' ')returns the array
$a
with 5 members, the strings
"This", "is", "a", "valuable", "string"
.
The method may be used in:
$env:path.split(';')which returns the Windows environmental variable
path
with
its entries stacked vertically.
-eq, -ne, -gt, -ge, -lt, -le, -like, -notlike, -match, -notmatch,
and -cmatch
. Although the operator -replace
does not perform a comparison,
it is usually included in this group.
'peanutbutter' -like 'nut' # false 'peanutbutter' -like '*nut*' # true (* is wildcard) 'peanutbutter' -notlike '*nut*' # false 'peanutbutter' -notlike 'nut' # true 'peanutbutter' -match '[a-z]+' # true (regexp: all letters) 'peanutbutter' -match 'r$' # true (regexp: last letter is r) 'peanutbutter' -match '[A-Z]+' # true (PS is case insensitive) 'peanutbutter' -cmatch '[A-Z]+' # false (cmatch matches cases) 'peanutbutter' -replace 'u', 'U' # 'peanUtbUtter'A website may contain the microsoft installation app
UsefulApp.msi
and its SHA256 hash
in the text file UsefulApp.msi.digest
(64 hex digits). Download both and issue:
(Get-FileHash .\UsefulApp.msi).Hash -eq (cat .\UsefulApp.msi.digest)[0]If this returns
True
, you can trust that the downloads haven't been changed on their way down to your computer.
The cmdlet Get-FileHash
returns an object with 3 members, one of them called Hash
.
Because cat
returns an array, you need to pick out its first element.
Note parenthetically that when the publisher of UsefulApp.msi
uses MicroSoft Authenticode, the cdmdlet Get-AuthenticodeSignature
may be worth looking at.
switch
statement is case sensitive because of the flag -casesensitive
:
function f ($str) { Switch -casesensitive ($str) { 'aap' { write-host 'AAP' } 'noot' { write-host 'NOOT' } 'mies' { write-host 'MIES' } 'wim' { write-host 'WIM' } Default { "Unable to determine value of $str" } } "Statement after switch" } f('noot') # --> NOOT \n Statement after switch f('Noot') # Unable to determine value of Noot \n Statement after switch
[console]
with methods (among others) beep
and write
:
see
the msdn (microsoft developer network) site.
[console]::beep(800, 1000) # beep at 800 Hz for 1000 msec [console]::write([char]0x07) # Write hex 07, that is, ring the bell (does not work under ISE) [console]::readkey() # Return name of key + modifier(not under ISE)Another built-in class is
[math]
. See
msdn.
[math]::pi # 3.14159265358979 [math]::cos([math]::pi) # -1 [math]::max(-1, -4) # -1 [math]::pow(10,3) # 1000
Get-Random -min 0.0 -max 1.0 # Random nr between 0.0 and 1.0, see help Get-Random for boundsAlternatively, use the System.Random object:
$rand = New-Object Random $rand | gm # Gives methods of $rand, among which NextBytes
[byte[]]$out = @(0,0,0,0) <--> $out = New-Object Byte[] 4Fill the array by a method of instance
$rand
:
$rand.NextBytes($out) # Fill array $out with 4 integer random numbers n: 0 -le n -le 255 $out # To console
-ErrorAction
, abbreviated: -ea
. The parameters
of this flag (Continue
, etc) may be replaced by numbers, as follows:
# -ErrorAction Continue | Ignore | Inquire | SilentlyContinue | Stop # -ea 2 | 4 | 3 | 0 | 1For instance, suppress error message about inaccessible subdirectories as follows:
ls -rec -ea 0 *.jpg # all .jpg files in present and all subdirectories
As in many languages errors may be trapped. Enter Get-Help about_trap
to see how.
The very same info as web page is here:
About trap.
Example of trapping:
Trap [System.Exception] { "Command error trapped.`n$_" # Automatic variable '$_' contains system error msg; `n gives newline. continue # Suppress traceback, continue after erroneous statement } nonsenseString # Erroneous statement: unknown cmdlet, function or script. 'Execution continues'PS also has a
Try ... Catch
construct:
try { An error # Illegal statement } catch { "An error occurred" }The global object
$error
is a stack containing the consecutive non-trapped errors. To list the latest
and the first error, respectively:
$error[0] | fl # The latest $error[$error.count-1] | fl # The first
C#
("C sharp") and is an intrinsic part of PS. A function name, just like a cmdlet name, is preferably
of the form "verb-noun" where "verb" is any of the existing verbs.
gcm -commandtype function |select verb -unique |sort verb|ft
Example of a user function:
function Write-ToScreen($path, $name) { Write-Host $path, $name }Alternatively, the parameters can be defined by the
param
statement:
function Write-ToScreen { param($path, $name) Write-Host $path, $name }Both forms can be called as (parameters can be abbreviated to unique strings):
Write-ToScreen -path roaming -name debug.txt # --> roaming \n debug.txtOr more briefly:
Write-ToScreen roaming debug.txt # Quotes are not requiredOr more classically:
Write-ToScreen('roaming', 'debug.txt') # Quotes are required
Write-Output
.
function list-no{ write-output "first" "second" "third" |fl } $a = list-no # Nothing to the console; output collected in return array $a $a # --> first \nsecond \nthird (vertically stacked)Not all console output is collected: the cmdlets
Write-Host
and Out-Host -i
write only to the console (do not generate return values):
function list-yes{ Write-Host "first" Out-Host -i "second" } $a = list-yes # --> first \nsecond (vertically stacked) to console $a # No outputBecause the return array may be redirected to a file and not all console output generated during function execution ends up in this array, the return values require close inspection. Example:
function Write-Vars { $a = 'A'; $b = 'B'; $c = 'C'; $d = 'D'; $e = 'E'; Write-Output $a # To return array (including EOL chars) $e | fl # To return array Write-Host $b # To console $c # To return array Out-Host -i $d # To console } $f = Write-Vars # 'B', 'D' to console; 'A', 'E', 'C' to $f $f # 'A', 'E', 'C' to console Write-Vars # 'A', 'E', 'B', 'C', 'D' to console (in order of assignment). Write-Vars > foo.out # 'B', 'D' to console; 'A', 'E', 'C' to foo.outOne could expect that the statement
Write-Vars
would write first the immediate values
of $b
and $d
followed by the values of the return array, but this is not the case.
function list-txt{ls *.txt} $a = list-txt # No console output, output of ls returned. $a # Info of .txt files to console list-txt | ft name # Only file names to console function list-txt{ls *.txt|write-host} $a = list-txt # Long file names of .txt files to console, nothing to $a. $a # No output, no returned parameters.
Function parameters can have a default value:
function f { param($c = "Pete") $c } f # -> Pete f John # -> JohnPositional parameters are passed in the automatic array
$args
:
function Get-Pos { foreach ($p in $args) { Write-Host $p } } Get-Pos 1, 2, 'pink elephant', icecream # --> 1 2 pink elephant icecreamFunctions may handle piped input (property names must be known—as always in pipelines). The named members of an object or hash table piped into the function are handled in a
Process block
:
Function Test-PipedValue { Process { Write-Host "name: " $_.name, "color: " $_.color } }Example, create hash table and pipe into function:
$hash = @{name = 'Jean'; color = 'White'} $hash | Test-PipedValue # --> name: Jean color: WhiteThe process command loops over elements of the piped object:
function Test-Loop{ process { Write $_ # alias of write-output, appends EOL char } } 1, 2, 3, 4 | Test-Loop # --> 1\n 2\n 3\n \4See
help about_functions
for more info.
.ps1
.
The script is invoked from a PS session by entering its file name (= script name) prefixed by its path.
If the script has parameters (defined in a param
statement) their values follow the script name,
optionally prefixed by the parameter names, just as is the case in a function invocation.
The output of scripts is comparable to that of functions. That is,
the command Write-Host
writes immediately (and only) to the console, while Write-Output
writes to a return array. After termination of the script it is decided what happens to the return array:
redirected to a file or to the console.
Create the script Measure-Text.ps1
containing the following 8
lines:
# Begin Measure-Text.ps1 Param ([string]$filename) # Script with one parameter, a string. Write-Output "`nStatistics of $filename `:" # To return array cat $filename | measure -line -word -character|ft # To return array, counts of lines, words, and chars Write-Host "You will hear a beep after 2 sec:" # To console sleep -sec 2 Write-Host 'Beep'; [console]::beep(800,1000); # To console # End Measure-Text.ps1Compare what is written on the console when the output is redirected (the script measures its own length):
.\Measure-Txt.ps1 -filename .\Measure-Txt.ps1 > foo.out cat foo.outto when the script is called directly (parameter name
-filename
is optional and omitted):
.\Measure-Txt.ps1 .\Measure-Txt.ps1Variables and functions have a default scope which may be modified. For example, inside a script the following function has only the script as scope. The scope of the variable
$a
is modified to global and hence
$a
is known to the invoking PS session:
function Display-Hello { "Hello, World" $global:a = 13 } Display-Hello $a
HKCU
and HKLM
.
The hives
HKCR, HKU
, and HKCC
are not directly accessible by cd
. One must issue
cd registry::hkcr
to access HKCR:
, etc. The latter change of directory leads to the very long prompt:
PS Microsoft.PowerShell.Core\Registry::HKCR>Alternatively, one can define a new PSdrive by a command like:
New-PSDrive -PSProvider registry -Root HKEY_CLASSES_ROOT -Name HKCRfollowed by
cd HKCR:
. The advantage is the much shorter prompt string: PS HKCR:\>
.
set ps 'PSChildName' -opt constantso that from hereon
$ps -eq 'PSChildName'
.
In the examples below the current PSdrive is HKCU\software
. One gets there in
a PowerShell session by issuing
cd hkcu:\softwareRecall that
-ea 4
stands for -ErrorAction Ignore
. This
flag suppresses errors about non-accessible keys of which there are many in the Registry.
The alias gp
stands for Get-ItemProperty
.
The alias gi
stands for Get-Item
. The functionality of gi
overlaps to a large extent with ls
and gp
.
Now follows a list of examples that may be useful in inspecting/traversing the Registry:
ls -rec -depth 1 -name # Tabulate names of subkeys (children) and subsubkeys (grandchildren). gp * # Lists all value entries (name and value) in all subkeys plus a few PS variables that define # three generations of the current path. Skips empty subkeys. ls 7-zip | gp # Value entries (names and values) of the subkeys of 7-zip (plus a few PS variables). gp . # Lists value entries in present key. No output if present key is empty. gp * |ft $ps # $ps='PSChildname' is a PS variable, returned by gp, that contains the name of subkey (child). # Hence this tabulates names of all non-empty subkeys. gp microsoft # No output because no value entry present in subkey 'microsoft' (is empty). gp RegisteredApplications # Value entries (names and values) of hkcu:\software\RegisteredApplications + PS info gi * # Same as ls *; lists names of subkeys (also empty ones) and their value entries; # no output when present key is endnode. gi . # Value entries of present key. Almost same info as 'gp .', but somewhat different format. gi .\Microsoft\Notepad # Value entries of subkey gi .\microsoft\Notepad |fl * # A subkey is an object, list its property names and values. One member is array 'Property' gi .\microsoft\Notepad |select -exp property # The content of array 'Property' in expanded form, # the array elements contain the names of value entries of the present key.Note on ls * −rec −depth 1
HKCU:\software
:
ls * -rec -depth 1 -ea 4 | measure -line # gives 35329 lines ls -rec -depth 1 -ea 4 | measure -line # gives 804 lineswhile both commands—issued from
c:\
—give the very same number of lines (542).
It is difficult to see this dependence on context as anything but a bug. It is, therefore, advisable to never use ls *
together with the flags -rec -depth
.
ls -rec -depth 1 |select name, propertyHere
property
contains the names of all value entries, but only the first few elements are listed.
To expand this array, together with the names of the subkeys separated by an empty line, use the following:
ls -rec -depth 1 |select name, property |% {$_.name; $_.property; "" }Explanation:
ls
returns an array of subkey objects that all have the properties name
and property
. The cmdlet select
picks from each subkey object these
two properties and adds them to an object that is referred to by
$_
in the next stage of the pipeline.
These objects are collected in an array that is passed to % = Foreach-Object
. Then %
loops over the array elements.
The script block in the body of the loop simply issues the two member names as commands.
Issuing of a variable name gives the writing of the content of the variable. If the content is an array,
the array is written element for element, every element on a new line.
Set-Location
(alias cd
) enables this.
The -recurse
parameter of ls
does not imply any cd
in a script block.
An explicit cd
is necessary if some processing must be performed lower down the path.
For example, recalling that ls -name -rec -de 1
returns an array containing names of subkeys and subsubkeys,
we see that the following statement gives the names of all subkeys and subsubkeys
by execution of pwd
(which returns the name of the present key),
including those without value entries (the empty ones):
ls -name -rec -depth 1 |% {$p=pwd; cd $_ -ea 4; pwd; cd $p; rv p}Here
-ea 4
suppresses the listing of an error when a cd
to a non-accessible subkey is attempted.
The newly created variable
$p
is removed to avoid possible later side effects.
Compare this statement to the following command that lists names (contained in PSPath
) of
only non-empty subkeys and subsubkeys, but does not require a cd
:
ls -name -rec -depth 1 |gp |select pspathOnce one has descended to a certain key, the names of its value entries (if any) are obtained by:
gi . |select -exp property # This returns line by line the names of the value entries.The command
gpv -name entry_name
returns the value coupled to entry_name
. For instance,
the subkey 7-zip
of HKCU\software
contains the value entry pair (lang: nl)
,
i.e., it has entry name lang
and entry value nl
.
The commands issued from HKCU\software
:
$p=pwd; cd 7-zip; gpv -name lang; cd $p; rv p;effectively leave us in
HKCU\software
and outputs the entry value nl
named lang
.
gi . |select -exp property |% {$v=gpv -name $_; write-host $_":", $v; rv v;}The next command lists names (
$p
) of non-empty subkeys, subsubkeys, and subsubsubkeys,
name of value entry ($name
) and corresponding value ($v
).
It does not list the value entries of the current key.
Note the nesting of |%
and the line continuation:
ls -rec -name -de 2 |% {$cd=pwd; $p=$_; cd $p -ea 4; gi . |select -exp property|` % {$name=$_; $v=gpv -name $name; write-host $p": ", $name" = "$v; rv name, v }; cd $cd; rv cd, p}The final command finds entry values containing a given string in keys below the current key. It is important to note that the search time increases exponentially with the value of the parameter
-depth
:
$string = "aul" ls -rec -name -depth 2 |% {$cd=pwd; $p=$_; cd $p -ea 4; gi . |select -exp property|` % {$name=$_; $v=gpv -name $name; if ($v -match $string) {write-host $p": ", $name" = "$v;}; rv name, v }; cd $cd; rv cd, p} rv string